1

Test Report for Hermes Lite SDR

June 14-17, 2015

by Adam Farson VA7OJ/AB4OJ (Iss. 3)

Figure 1: The Hermes Lite.

Introduction: The following are the results obtained from receiver and transmitter tests which the author conducted on a Hermes Lite direct-sampling/DUC SDR receiver-exciter, kindly loaned by Dave Miller VE7PKE. The DUT was connected via a dedicated Ethernet router to a 2.3 GHz Core i5 laptop running under Win7 Pro 64-bit. As the Hermes Lite does not have an on-board DAC, VAC (Virtual Audio Cable) was enabled in PowerSDR. Software version: OpenHPSDR mRX PS V3.2.27. Firmware version: Hermes 3.1. Initial DUT configuration: https://github.com/softerhardware/Hermes-Lite/wiki/Software#hpsdr Note that the Dither and Random checkboxes have been re-purposed. Dither OFF increases LNA gain by 32 dB. Random ON enables a unique AGC function controlled by a clipping detector. For these tests, Enable Attenuator (Options tab) is ON, enabling S-ATT. In the HPSDR tab, Dither is ON and Random OFF (the latter to allow measurement of ADC clip level). S-ATT is set at 0 dB. This sets LNA gain at +19 dB with Dither ON and +48 dB with Dither OFF. The +19 dB LNA gain value was selected because it yields MDS in the -120 dBm range, comparable to other direct-sampling SDR receivers.

A. Receiver Tests: 1. MDS (Minimum Discernible Signal). Marconi 2018A signal generator, 20 dB pad, DUT ANT1

input. Ballantine 323 true RMS voltmeter at laptop PHONES jack.

2. NPR (noise power ratio). Wandel & Goltermann RS-50 and RS-25 noise generators (fitted with filters per Table 2), MCL 75/50Ω transformer, DUT ANT1 input. DUT set to LSB for all test frequencies except 5340 kHz (USB). 2.4 kHz channel filter selected. NPR read directly from spectrum scope. (Ref.2).

2

3. RMDR (reciprocal mixing dynamic range). Marconi 2018A signal generator, 3 dB pad, 9.83 MHz

bandstop filter, 0-110 dB step attenuator, DUT ANT1 port. Ballantine 323 true RMS voltmeter at laptop PHONES jack.

4. 2-Tone 3rd-Order IMD (IFSS). Marconi 2018A and 2019 signal generators, MCL ZHL-32A buffer amplifiers, MCL low-pass filters appropriate for band tested, 20 dB pads, MCL ZSC-2-1W combiner, 0-110 dB step attenuator, 10 dB fixed attenuator at DUT ANT1 input. IMD product levels read from S-meter. Test results presented as curves of IMD level vs. test signal power, with ITU-R P.372-1 band noise levels as datum lines. (Ref.5, Slides 23-27.)

5. DR2 (IMD2 dynamic range). As for IFSS.

6. NR SINAD Improvement. As for MDS. HP 339A distortion meter at laptop PHONES jack.

7. NB impulse response. HP 8011A pulse generator connected to DUT ANT1 input. Headphones at laptop PHONES jack.

8. ANF tone suppression. As for MDS, HP 3580A spectrum analyser at laptop PHONES jack.

9. APF passband. As for MDS, HP 3580A spectrum analyser at laptop PHONES jack.

1. MDS (Minimum Discernible Signal tested in CW mode (B = 500 Hz), S-ATT = 0 dB, LNA gain +19

and +48 dB. The receiver is tuned to each test frequency f0 in turn and a test signal applied. At each frequency, the input power Pi required to raise audio output noise level by 3 dB is noted. Pi

= MDS. Test Conditions: NR/NB/ANF off, S-ATT = 0 dB, AGC Slow, AGC Gain at max. AVG on, display refresh rate 60 fps. See Table 1.

Table 1: Minimum Discernible Signal (MDS).

f0 MHz MDS dBm

LNA Gain dB +19 +48

3.6 -119 -129

14.1 -118 -128

28.1 -116 -126

2. NPR (Noise Power Ratio), tested in SSB mode (B = 2.4 kHz. Receiver tuned to notch center in all

cases. Noise loading set to the point where ADC just does not clip for 10 sec. (-1 dBFS). NPR read off spectrum scope (noise level in a channel outside notch minus noise level at bottom of notch.) See Table 2 and Figure 2.

Test Conditions: Receiver tuned to bandstop filter centre freq. f0 ± 1.5 kHz, SSB, B = 2.4 kHz, S-ATT = 0 dB, LNA gain +19 dB, NR off, NB off, ANF off, AGC slow, AGC Gain at max., AVG on, Display Refresh Rate 60 fps.

Table 2: NPR Test Results.

DUT Det BW kHz Mode BSF kHz BLF kHz Equiv. J3E chnls. PTOT dBmc NPR dBa Theor. NPR dBb

Hermes Lite 2.4

LSB 1940 60…2044 480 -28 57 60.8

LSB 3886 60…4100 960 -29 59 57.7

USB 5340 60…5600 1260 -30 57 56.3

LSB 7600 316…8160 1800 -31 56 54.8

3

Notes on NPR test:

a. NPR read directly off spectrum scope. b. Theoretical NPR is calculated for the 12-bit ADI AD9886 ADC using the method outlined in Ref.1,

normalised to 12 bits. The theoretical NPR value assumes that BRF is not limited by any filtering in the DUT ahead of the ADC, and that the net gain between the antenna port and the ADC is 0 dB.

c. S-ATT = 20 dB increases PTOT for same NPR reading by 20 dB.

Figure 2: NPR at 5340 kHz.

3. RMDR (Reciprocal mixing dynamic range). Marconi 2018A signal generator, 3 dB pad, 9.830 MHz 4-pole bandstop filter (> 80 dB notch depth), 0-110 dB step attenuator, DUT (RX port). Noise floor read on S-meter in CW mode (500 Hz bandwidth) with DUT terminated in 50Ω. S-ATT = 0 dB, LNA gain = +19 dB, NB/NR/ANF OFF. Signal generator frequency (f0) tuned for max. null; DUT tuned to f0. Signal generator then tuned to f0 - offset and output Pi increased to raise detected noise by 3 dB. RMDR = Pi – (noise floor with generator off.) See Table 3. Note: The residual phase noise of the measuring system is the limiting factor in measurement accuracy.

Table 3: Reciprocal Mixing Dynamic Range (RMDR) at 9830.28 kHz.

Δf kHz Pi dBm RMDR dB Phase noise dBc/Hz

1 +4.4 116 -143

2 > +2.4 > 117 -144

3 > +2.4 > 117 -144

5 > +2.4 > 117 -144

10 > +2.4 > 117 -144

Noise Floor = -120 dBm

4. Two-Tone IMD (IFSS, Interference-Free Signal Strength) tested in CW mode (500 Hz), S-ATT = 0 dB, LNA gain = +19 dB. Test frequencies per Table 4. Absolute IMD product level is read on S-meter in a 500 Hz CW detection bandwidth at various test-signal power levels with S-ATT = 0 dB. The ITU-R P372.1 band noise levels for typical urban and rural environments are shown as datum lines. Figure 2 is a typical screenshot of the IFSS test. The results are given in Figure 4.

Table 4: IFSS Test Frequencies & IMD Products.

Band f1 kHz f2 kHz Lower IMP kHz Upper IMP kHz

20m 14100 14102 14098 14104

4

Figure 3. Typical IFSS spectrum scope display, showing IMD3 products.

Figure 4: 20m IFSS (two-tone IMD) curve for S-ATT = 0 dB.

Note on IFSS (2-tone IMD3) test: This is a new data presentation format in which the amplitude

relationship of the actual IMD products to typical band-noise levels (Ref.3) is shown, rather than

the more traditional DR3 (3rd-order IMD dynamic range). The reason for this is that for an ADC,

SFDR referred to input power rises with increasing input level, reaching a well-defined peak and

then falling off. In a conventional receiver, SFDR falls with increasing input power. See also

Ref.5, Slides 23 – 27.

5

The SFDR (spurious-free dynamic range) behaviour of an ADC invalidates the traditional DR3

test for a direct-sampling SDR receiver. Our goal here is to find an approach to SFDR testing

which holds equally for SDR and legacy receiver architecture.

It will be seen from the above chart that the IMD curve approximates 1st order until -10 dBFS (10

dB below ADC clip level) is approached; in Figure 4, the curve in the range -10 dBFS to 0 dBFS is

closer to 3rd-order. This is due to non-linearity in the LNA.

5. DR2 (IMD2 dynamic range) tested at 14.200 MHz, WIDE, in CW mode (500 Hz), S-ATT = 0 dB, LNA gain = +19 dB. Test frequencies: f1 = 6100 kHz, f2 = 8100 kHz. 2nd-order IMD2 product: 14200 kHz. Test-signal level is adjusted for a 3 dB increase in audio output, and DR2 & IP2 calculated. DR2 = Pi – MDS. IP2 = (2 * DR2) + MDS Test Results: Refer to Table 5.

Table 5: IMD2 Dynamic Range (DR2).

S-ATT dB LNA Gain dB MDS dBm Pi dBm/tone DR2 dB IP2 dBm

0 +19 -118 -75 43 -32

6. Noise Reduction (NR) SINAD Improvement tested in USB mode (2.4 kHz), ), S-ATT = 0 dB, LNA gain = +19 dB. Receiver tuned to 14100 kHz. Test signal at 14101 kHz. The distortion meter is connected to the PHONES jack and signal level adjusted for 6 dB SINAD with NR off. NR and NR2 are selected in turn, and SINAD read and noted for each setting. PowerSDR NR parameters are at default values.

Test Results: Refer to Table 6.

Table 6: NR Improvement.

NR Setting SINAD dB SINAD Improvement dB

OFF 6 0

NR 13 7

NR2 18 12

Note: With NR2 on, the recovered audio has a watery, burbling quality.

7. Noise Blanker (NB) Impulse Suppression tested in USB mode (2.4 kHz), S-ATT = 0 dB, LNA gain = +19 dB. A pulse train is applied to DUT RX port. Receiver tuned to 3600 kHz. Pulse parameters: Vpk = 32 mV, rise-time ≈ 10 ns, PRF = 2 pps. Test Results: With NB on, PowerSDR NB parameters at default values, no audible pulse “ticks”

with duration ≤ 225 ns. Noise floor decreases slightly with NB on.

8. Auto-Notch Filter (ANF) Tone Suppression tested in USB mode (2.4 kHz), S-ATT = 0 dB, LNA gain = +19 dB, PowerSDR NB parameters at default values. The DUT is tuned to 14100 kHz, and a S9 + 20 dB (-50 dBm) test signal at 14101 kHz is applied. The test tone level at the PHONES jack is measured with the HP 8530A spectrum analyser.

Test Results: ANF reduces the tone level by more than 70 dB.

6

9. Audio Peak Filter (APF), tested in CW mode (B = 500 Hz), S=ATT = 0 dB, LNA gain = +19 dB. The

HP 3580A spectrum analyser is connected to the laptop PHONES jack. A test signal at 14.1 MHz and -90 dBm is applied to RX IN. The spectrum analyser is tuned to f0 = 600 Hz and screenshots taken for various APF settings. See Figures 5, 6 and 7. HP 3580A settings: 10 Hz RBW, 50Hz/div span, 2s/div sweep time, ref. level 0 dBv.

Figure 5: APF off.

Figure 6: APF 75 Hz, 0 dB.

7

Figure 7: APF 75 Hz, +10 dB.

Figure 8: APF 30 Hz +10 dB

8

B. Transmitter Tests Attenuators used in these tests should be rated at 2W or higher.

1. Maximum RF output. HP 432A power meter with Harris N6284B sensor connected to DUT TX

port via 20 dB attenuator. TUNE mode selected, Drive at 100%. Test Results: See Table 7.

Table 7: Max. RF output PO.

f0 MHz Max. PO dBm

3.6 +12

14.1 +15

28.2 +18.5

Note: PO checked with SSB, Mode = Radio, Level = 0.0 dB, Freq. = 1000 Hz selected in Transmit. P0 readings identical to TUNE mode.

2. SSB 2-Tone Transmitter IMD. In PowerSDR Tests tab, Freq. 1 = 700 Hz, Freq. 2 = 1700 Hz, Level = 0 dB, RF Power = 100%. HP 8563E spectrum analyser connected to DUT TX port via 30 dB attenuator. IMD measured on 3.6, 14.1 and 28.1 MHz. Refer to Table 9, and Figures 9 - 11.

Table 9: TX IMD in dBc (ref. 1 of 2 equal tones) at 100% RF Power.

Freq. MHz 3.6 14.1 28.1

PEP dBm +12 +15 +18

IMD3 -54 -48 -52

IMD5 -72 -65 -66

IMD7 -84 -77 -75

IMD9 < -86 < -86 < -90

Subtract 6 dB for IMD ref. 2-tone PEP.

3. SSB TX Noise IMD. As for Test 2, except Mode = Noise, Level = 0.0 dB selected in Transmit. Test

run at 14.1 MHz, P0 = 100%. Power in noise bandwidth ≈ +5 dBm. See Figure 12.

9

Figure 9.

Figure 10.

10

Figure 11.

Figure 12.

11

4. TX harmonics & spurs. HP 8563E spectrum analyzer with HP 85672A spurious response utility, connected to DUT ANT1 via 30 dB attenuator. Test run in SSB mode at PO = 100% on 3.6, 14.1 and 50.1 MHz.

Test Results: Refer to Figures 13 – 15 (harmonic charts) and 16 – 18 (sweeps, showing harmonics and non-harmonic spurs).

Figure 13.

12

Figure 14.

Figure 15.

13

Figure 16.

Figure 17.

14

Figure 18.

5. TX harmonics & spurs with harmonic filters. At the suggestion of Steve Haynal KF7O, the

author repeated Test 4 with a Mini-Circuits (MCL) LPF inserted between DUT TX OUT and the

attenuator ahead of the spectrum analyser. The appropriate filter was selected for each band in

turn.

HP 8563E spectrum analyzer with HP 85672A spurious response utility, connected to DUT ANT1 via 30 dB attenuator. Test run in SSB mode at PO = 100% on 3.6, 14.1 and 50.1 MHz.

Test Results: Refer to Figures 19 – 21 (harmonic charts) and 22 – 24 (sweeps, showing harmonics and non-harmonic spurs). Table 10 gives the LPF part number for each band tested.

Table 10: MCL LPF Part Numbers

Band m TX Freq. MHz MCL LPF Part No.

80 3.6 BLP-5

20 14.1 BLP-15

10 28.1 BLP-30

15

Figure 19.

Figure 20.

16

Figure 21.

Figure 21.

17

Figure 22.

Figure 23.

18

C. Comments:

1. The receiver’s DR2/IP2 (2nd-order IMD) performance is inadequate for use without a good preselector in Region 1 or in other areas where high-power HFBC transmitters are present.

2. As the Hermes Lite FPGA code does not support CW generation, no CW tests were conducted.

3. The default receiver front-end configuration (Dither ON, Random OFF, S-ATT = 0 dB, yielding +19

dB LNA gain) was used in all receiver tests unless otherwise specified, so as to ensure maximum usable dynamic range. In addition, the NPR and IFSS tests require that the onset of ADC clipping will not alter RF gain.

4. The transmitter’s excellent linearity makes it suitable as an exciter for a wide variety of PA chains. It should be noted though, that harmonic filters are required to meet regulatory requirements. This is especially true on 10m, where spurs at -10 and -20 dBc were observed.

5. The additional tests with harmonic filters at the DUT TX OUT port show excellent harmonic suppression, although on the 10m band a significant spur (-20 dBc) at 35 MHz is still visible.

D. Acknowledgements:

I would like to thank my friend Dave Miller VE7PKE for the loan of the Hermes Lite DUT, and also for guiding me through the on-line Hermes Lite literature.

E. References:

1. http://www.ab4oj.com/test/docs/16bit_npr.pdf (theoretical maximum NPR) 2. http://www.ab4oj.com/test/docs/npr_test.pdf (NPR testing) 3. http://www.ab4oj.com/icom/nf.html (antenna & receiver NF) 4. http://www.nsarc.ca/hf/arrl_test.pdf (brief tutorial on radio testing) 5. http://www.nsarc.ca/hf/rcvrtest.pdf (IFSS: Slides 23 – 27)

Adam Farson VA7OJ/AB4OJ, June 17, 2015. Copyright © 2015 A. Farson VA7OJ/AB4OJ. All rights reserved.